SF2A 2011 G. Alecian, K. Belkacem, R. Samadi and D. Valls-Gabaud (eds) DEVELOPMENT OF TECHNIQUES TO STUDY THE DYNAMIC OF HIGHLY ELLIPTICAL ORBITS G. Lion1 and G. M´etris1 Abstract. Many spacecrafts are or will be placed in highly eccentric orbits around telluric planets of the Solar system. Such eccentricities allow to cover a wide range of altitudes, mainly for planetology purposes. There are also orbits with very high eccentricity around the Earth, especially the GTO (Geostationary Transfer Orbit) and orbits of some space debris. In this case, the motion is strongly perturbed by the luni- solar attraction. For various reasons which will be recalled, the traditional tools of celestial mechanics are not well adapted to the particular dynamic of highly eccentric orbits. Therefore, it is necessary to develop specific techniques for this configuration. This concerns numerical as well as analytical tools. We will show how to construct the expression of the disturbing function due to the presence of an external body, well- suited for highly eccentric orbits. Expansion of the elliptic motion in closed-form by using Fourier series in multiple of the eccentric anomaly will be presented. On the other hand, classical methods of numerical integration have often a poor efficiency. We will show the interest of geometric integrators and in particular the variational integrators. Keywords: Third-body, disturbing function, Hansen-like coefficients, elliptic motion, high eccentricity, closed-form, variational integrators 1 Introduction When dealing with highly elliptical orbits, we have to face several difficulties. Due to the fact that such orbits cover a wide range of altitudes, the hierarchy of the perturbations acting on the satellite changes with the position on the orbit. At low altitude, the oblateness of the Earth (the so called J2 effect) is the dominant perturbation while at high altitude the luni-solar perturbation acceleration can reach or exceed the order the J2 acceleration. This particular configuration requires to develop adapted strategies to propagate the orbit by means of analytical theories form the one hand and numerical integration on the other hand. From the analytical point of view, the traditional theories of celestial mechanics are not well adapted to this particular dynamic. On the one hand, analytical solutions are quite generally expanded into power series of the eccentricity and so limited to quasi-circular orbits. On the other hand, the time-dependency due to the motion of the third body is almost always neglected. Regarding the numerical methods, the traditional integrators can be numerically unstable for high eccen- tricity if a moderate step size is chosen due to the very fast variation of the perturbation around the perigee. If the step size is taken extremely small this implies large round-off errors and hight CPU cost. Experiments show that even numerical integrators with variable step size does nit solve perfectly this problem. The paper is organized as follows. In Section 2, we propose a new expression of the disturbing function of the third-body problem which is in closed form with respect to the satellite eccentricity and still permits to construct an analytical theory of the motion. We will show that the use of the eccentric anomaly instead of the mean anomaly as fast angular variable fulfills this requirement. In Section 3, we give an overview of the variational integrators and we will see the interest of using such methods rather than classical integrators for orbital mechanics problems. 1 Universit´ede Nice Sophia-Antipolis, Centre National de la Recherche Scientifique (UMR 6526), Observatoire de la C^ote d'Azur, G´eoazur, Avenue Nicolas Copernic 06130 Grasse, France c Soci´et´eFrancaise d'Astronomie et d'Astrophysique (SF2A) 2011 674 SF2A 2011 2 Third-body problem 2.1 Expression of the disturbing function of the third-body problem Let us consider a satellite of position vector r = ru orbiting a central body and a third body of position vector r0 = r0u0, with u and u0 unit vectors. The disturbing function R of the third-body problem can be expressed into spherical coordinates (r; φ, λ) using the traditional expansion in Legendre polynomials Pn(x) as follow µ0 X r n R = P (u · u0) ; (2.1) r0 r0 n n≥2 where µ0 = Gm0, m0 being the mass of the third body and G the gravitational constant. In order to construct an analytical theory, it is more suitable to express (2.1) as function of orbital elements (semi-major axis a, eccentricity e, inclination I, argument of perigee !, longitude of the ascending node Ω and mean anomaly M) or equivalent variables. From several works as Kaula (1962), Giacaglia & Burˇsa(1980), Lane (1989) or yet Brumberg (1995), we were able to obtain a general expression of the disturbing function into Hill-Whittaker canonical variables: r,_r, θ = ! + ν, G = pµa(1 − e2), Ω and H = G cos I, with ν the true anomaly: 0 n n n n n n 0 n+1 µ X X X X X a r a 0 R = D 0 0 exp i Ψ − Ψ 0 0 ; (2.2a) a0 a0 a r0 n;m;m ;p;p n;m;p n;m ;p n≥2 m=−n m0=−n p=0 p0=0 0 0 m−m0 (n − m )! 0 Dn;m;m0;p;p0 (I;I ;") = (−1) Fen;m;p(I)Fen;m;0p0 (I )Un;m;m0 (") ; (2.2b) (n + m)! Ψn;m;p = (n − 2p)θ + mΩ ; (2.2c) 0 0 0 0 0 Ψn;m0;p0 = (n − 2p )θ + m Ω ; (2.2d) where the Fe-functions are related to the Kaula's inclination functions (see Kaula (1961)), " is the obliquity and the U-functions are to the Wigner formula (see Sneeuw (1992)) giving the components of the rotation matrix between equatorial to ecliptic plane related. In order to have a perturbation fully expressed in orbital elements, we expand the functions of the elliptical motion (r=a)n exp iν and (a0=r0)n+1 exp iν0 into Fourier series with respect to an angular variable and coefficients which depend of the eccentricity. Quite generally, these functions are expanded in multiples of the mean anomaly as follow (see for example Kaula (1962), Giacaglia (1974)) 1 r n X exp imν = Xn;m(e) exp isM ; (2.3) a s s=−∞ n;m where the Xs are the so-called Hansen coefficients. In the general case, the series (2.3) always converge as Fourier series but can converge rather slowly (see e.g Brumberg & Brumberg 1999). Only in the particular case where e is small, the convergence is fast thanks to the d'Alambert property which ensure that ejk−qj can n;k be factorized in Xq (e). That is why the method is reasonably efficient for most of the natural bodies (in particular the Sun and the Moon) but can fail for satellites moving on orbits with high eccentricities. In this case, Fourier series of the eccentric anomaly E, are much more efficient : 1 r n X exp imν = Zn;m exp isE ; (2.4) a s s=−∞ where the Z-functions are called the Hansen-like coefficients. Expressions of these coefficients are given in Brumberg & Fukushima (1994) and can be computed using recurrence relations (see Lion & M´etris (2011)). Using this development, we have the double advantage when 0 ≤ jmj ≤ n (which occurs in the third-body problem) that these coefficients admit a closed-form representation and that the sum (2.4) is exactly limited to s = ±n (coefficients are null for jsj > n). Using the Fourier series (2.3) and (2.4) we show in Lion et al. (2011) that the disturbing function R takes the form: n n n n n+1 +1 X X X X X X X a R = A 0 0 0 exp iΘ 0 0 0 ; (2.5a) r n;m;m ;p;p ;q;q n;m;m ;p;p ;q;q n≥2 m=−n m0=−n p=0 p0=0 q=−n−1 q0=−∞ Development of techniques to study the dynamic of highly elliptical orbits 675 0 n µ a 0 n+1;n−2p −n−1;n−2p0 0 A 0 0 0 = D 0 0 (I;I ;")Z (e) X 0 (e ) ; (2.5b) n;m;m ;p;p ;q;q a0 a0 n;m;m ;p;p q q 0 Θn;m;m0;p;p0;q;q0 = Ψe n;m;p;q − Ψe n;m0;p0;q0 ; (2.5c) Ψe n;m;p;q = qE + (n − 2p)! + mΩ ; (2.5d) 0 0 0 0 0 0 0 Ψe n;m0;p0;q0 = q M + (n − 2p )! + m Ω : (2.5e) 2.2 Lie transformations perturbation method The idea is to use a perturbative method based on the time-dependent Lie transform Deprit (1969) in order to obtain an approximated analytical solution of the third-body problem. Because the canonical variable h = ! is not ignorable and g is not automatically removed in the same time that the fast angle l = M after a canonical transformation (contrary to the J2 problem case), our initial Hamiltonian H0 of order 0 contains not only the keplerian energy, but also the secular part of the J2 problem. The disturbing function R belongs to the hamiltonian H1 of order 1. In that way, H0 depends of the three momenta L, G and H which will allow to eliminate the three conjugate angles l, g and h from the transformed hamiltonian. Next, we use the homological equation providing the Lie generator W and the new Hamiltonian K at first order. The new Hamiltonian K is taken such as it does not depend on any angular variable.